Brassinosteroids Regulate Dissociation of BKI1, a Negative Regulator of BRI1 Signaling, from the Plasma Membrane

See allHide authors and affiliations

Science  25 Aug 2006:
Vol. 313, Issue 5790, pp. 1118-1122
DOI: 10.1126/science.1127593


Brassinosteroids, the steroid hormones of plants, are perceived at the plasma membrane by a leucine-rich repeat receptor serine/threonine kinase called BRI1. We report a BRI1-interacting protein, BKI1, which is a negative regulator of brassinosteroid signaling. Brassinosteroids cause the rapid dissociation of BKI1–yellow fluorescent protein from the plasma membrane in a process that is dependent on BRI1-kinase. BKI1 is a substrate of BRI1 kinase and limits the interaction of BRI1 with its proposed coreceptor, BAK1, suggesting that BKI1 prevents the activation of BRI1.

There are more than 400 serine/threonine receptor–like kinases predicted in the Arabidopsis genome (1). BRI1, the major brassinosteroid receptor of Arabidopsis (24), has been studied using loss-of-function mutants, overexpression, and biochemical analyses to identify the activation and specificity of plant receptor–like kinases (5). Brassinosteroids control physiological and developmental processes such as stem elongation, vascular differentiation, seed size, fertility, flowering time, senescence, and resistance to biotic and abiotic stresses (2, 6, 7). Direct binding of brassinolide (BL), the most active brassinosteroid, to the extracellular domain of BRI1 activates a preformed homo-oligomer. Auto- or trans-phosphorylation of the C terminus of BRI1 then enhances kinase activity and the affinity of BRI1 for BAK1, its proposed coreceptor (811). A version of BRI1 lacking the 41 C-terminal amino acids is a more active receptor but cannot be fully activated, suggesting that other factors are also required to regulate BRI1 activity.

Downstream from BRI1 and BAK1, BIN2, a glycogen synthase kinase-3 family member (12), negatively regulates brassinosteroid signaling by phosphorylating members of a plant-specific family of transcriptional regulators, defined by the BES1 and BZR1 genes (1316). In the presence of brassinosteroids, BIN2 is inhibited by an unknown mechanism, leading to the dephosphorylation of BES1 and BZR1. Dephosphorylated BES1 and BZR1 then homodimerize or cooperate with other transcription factors, which allows DNA binding and regulation of hundreds of brassinosteroid-responsive genes (1517).

To investigate the signaling events between the plasma membrane and transcriptional responses, we searched for proteins that interact with BRI1 using yeast two-hybrid screens with a cDNA library from Arabidopsis shoot apical meristems. We repeatedly identified two proteins that interacted with the intracellular domains of wild-type or kinase-inactive BRI1: a transthyretin-like protein (TTL), which is a negative regulator of brassinosteroid-related plant growth (18), and an expressed protein of unknown function, At5g42750. We designated At5g42750 as BKI1 for BRI1 Kinase Inhibitor 1. A simple modular architecture research tool [(SMART),] predicts BKI1 to encode a protein of 337 amino acids with two separate Ser-rich domains and an Asn-rich region (Fig. 1A). BLAST searches of the predicted BKI1 amino acid sequence identified a similar gene in rice, as well as multiple expressed sequence tags (ESTs) from other angiosperms, which in several cases appear to contain the entire predicted coding region (Fig. 1B and table S1). The rice protein was previously reported to interact with the kinase domain of rice BRI1, although its function is unknown (19). No other similar sequences similarities were identified in other species, which suggests that BKI1 may be angiosperm-specific.

Fig. 1.

In vivo and in vitro interaction of BKI1 with BRI1's intracellular domain. (A) The predicted domain structure of AtBKI1. The sequences underlined indicate the region that interacts with BRI1's kinase domain. (B) Alignment of the deduced amino acid sequence of AtBKI1 with BKI1-like proteins from Oryza sativa (OsBKI1; AP005891.3), Medicago truncatula (MtBKI1; AC157645_3.1, identified in the database of the Institute for Genome Research), Gossypium raimondii (GrBKI1, a putative full-length gene assembled from ESTs CO071302.1, CO081346.1, CO074191.1, and CO081345.1), and Euphorbia esula (EeBKI1, a putative full-length gene assembled from ESTs DV136456.1, DV156616.1, DV139943.1, DV131013.1, and DV131013.1). (C) The carboxyl domain of BKI1 is necessary and sufficient to interact with BRI1's kinase domain in yeast. Residues of BKI1 present in the constructs were 1 to 252 (BKI1-ΔCT1), 1 to 299 (BKI1-ΔCT2), and 253 to 337 (BKI1-CT). (D) BKI1 specifically interacts with BRI1. BKI1 fused with GAL4-AD (BKI1-pEXAD502) specifically interacts with the intracellular domain of BRI1 (BRI1-KD) fused with GAL4-DB in yeast. (E) BKI1-6XHIS interacts with GST-BRI1-KD in vitro. In the pull-down product, the GST (left) or GST-BRI1-KD (right) was detected by antibody to GST, and the 35S-Met–labeled BKI1-6XHIS was detected by autoradiography. (F) BKI1-FLAG interacts with endogenous BRI1 in planta. 1, Col-0; and 2, BKI1-FLAG overexpression line. BRI1 and BKI1-FLAG were detected by immunoblot with antibody to BRI1 and antibody to FLAG, respectively. (G to J) pBKI1::GUS is ubiquitously expressed. GUS reporter gene expression was monitored in 2-week-old seedlings' leaves and shoot apices (G), hypocotyls (H), and roots (I), and in flowers of adult plants (J).

Sequence alignments indicated that the C-terminal domain of BKI1 is the most conserved region [about 32% identity in the C-terminal region (residues 253 to 337)]. The C terminus was both necessary and sufficient to bind the kinase domain of BRI1 (BRI1-KD) (Fig. 1C). BKI1 associated specifically with the kinase domain of BRI1 and not with TTL, BIN2, or kinase domains of other receptor-like kinases tested, including BAK1 and NIK1, another member of the BAK1 subfamily (Fig. 1D). BKI1 did not interact with CLV1, a leucine-rich repeat receptor-like kinase (LRR-RLK) involved in shoot apical meristem development (1), nor did it interact with BRI1's closest relatives, BRL1 and BRL3 (20) (fig. S1), indicating that the interaction of BKI1 with BRI1 is highly specific. Glutathione S-transferase (GST) pull-down experiments using GST-BRI1-KD and 35S-Met–labeled BKI1-6XHIS further indicated that BKI1 interacts with the kinase domain of BRI1 (Fig. 1E). Immunoprecipitation experiments confirmed that endogenous BRI1 interacted with a BKI1-FLAG fusion protein in vivo (Fig. 1F).

BKI1's function in brassinosteroid signaling was explored in several ways. First, we made transgenic plants harboring a β-glucuronidase (GUS) reporter gene expressed from the promoter of BKI1 to observe the expression pattern of BKI1 (Fig. 1, G to J) during development. BKI1 was expressed in leaves, petioles, shoot apices, hypocotyls, roots, and flowers, indicating that BKI1 and BRI1 are coexpressed in a number of tissues (21). To explore the function of BKI1 in BRI1 signaling, we created RNA interference (RNAi) lines to inhibit BKI1 RNA levels. Real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis indicated that the transcript level of BKI1 was significantly reduced (50% to 76%) in many RNAi lines, compared with a control line (Fig. 2A). The RNAi lines had longer hypocotyls than the control line grown in short days (control, 2.20 ± 0.09 mm; RNAi-2, 2.74 ± 0.08 mm; and RNAi-7, 3.09 ± 0.12 mm), and the levels of BKI1 transcripts were negatively correlated with hypocotyl length, suggesting that BKI1 represses brassinosteroid-related growth (Fig. 2B). In contrast, overexpression of BKI1 with either a FLAG or a yellow fluorescent protein (YFP) tag resulted in dwarf plants resembling plants harboring weak alleles of bri1. These plants had a smaller rosette (under both short-day and long-day conditions), reduced stature and petiole length, rounder rosette leaves, and delayed flowering compared with wild type (Fig. 2, C to F).

Fig. 2.

BKI1 is a negative regulator of brassinosteroid signaling. (A) The expression level of BKI1 in RNAi knockdown plants and a control line determined by qRT-PCR. (B) Suppression of BKI1 expression leads to longer hypocotyls. Seedlings were grown in short-day photoperiods for 7 days. At least 10 short-day-grown seedlings were measured. (A and B) The pBKI1::GUS line was used as a control. RNAi-2 and RNAi-7 were two independent RNAi knockdown lines for BKI1.(C to E) Phenotype of plants overexpressing BKI1-FLAG, with Col-0 (left) and a BKI1-FLAG line (right). (C) Short-day-grown seedlings. (D) Long-day-grown seedlings. (E) Long-day-grown adult plants. (F) Leaf morphology of seedlings grown in short days. Col-0 (right) and a BKI1-FLAG line (left). (G) Overexpression of BKI1-YFP leads to a reduced response (hypocotyl length) to BL. At least 20 short-day-grown seedlings were measured. (H) Overexpression of BKI1-YFP leads to an enhanced response to BRZ220. At least 20 dark-grown seedlings were measured. (I) Overexpression of BKI1-YFP results in reduced accumulation of dephosphorylated BES1. Immunoblots with antibody to BES1 show levels of phosphorylated BES1 (p-BES1) and dephosphorylated BES1. Bottom band, loading control. (J) Overexpression of BKI1-YFP alters the expression of BR-regulated genes. CPD and DWF4 are BR down-regulated genes, and Saur-AC1 is a BR up-regulated gene. In (A), (B), (G), (H), and (J), error bars indicate standard error.

To determine whether the dwarf phenotype was the result of altered brassinosteroid sensitivity, we measured the response of a line over-expressing a BKI1-YFP fusion protein to BL. In the absence of applied BL, overexpression of BKI1-YFP led to about a 40% reduction of hypocotyl length in the light (Fig. 2E), and plants were less sensitive to BL applications by a factor of at least 100 compared with wild type, whereas det2-1, a brassinosteroid biosynthetic mutant, was highly responsive to BL treatment (Fig. 2G). Conversely, the BKI1-YFP line and det2-1 were more sensitive to a brassinosteroid biosynthesis inhibitor, BRZ220 (22), as indicated by the lower concentration of BRZ 220 needed for the BKI1-YFP line (∼0.5 μM) to achieve 50% inhibition of hypocotyl length than for Columbia (Col-0) (∼2.0 μM) (Fig. 2H).

To determine whether the growth inhibitory effect caused by the overexpression of BKI1-YFP was due to an inhibition of BRI1 signaling, we tested the phosphorylation status of a downstream biochemical marker, BES1 (10), by immunoblot analysis. Without applied exogenous BL, a considerable amount of dephosphorylated BES1 was present in the wild type, whereas the amount of dephosphorylated BES1 in a BKI1 overexpression line was almost undetectable, which was similar to det2-1. Treatment with 0.1 μM BL for 1 hour strongly stimulated the accumulation of dephosphorylated BES1 in wild type and det2-1, but to a lesser extent in the BKI1-YFP line (Fig. 2I), suggesting that BRI1 signaling was suppressed by BKI1 overexpression.

To determine whether overexpression of BKI1 alters the expression of brassinosteroid-responsive genes, we measured the RNA levels of three brassinosteroid-regulated genes by qRT-PCR, including two down-regulated genes, CPD and DWF4, and an up-regulated gene, Saur-AC1. As shown in Fig. 2J, compared with wild type, overexpression of BKI1-YFP led to a significant increase of the expression of CPD and DWF4 in the absence of BL (CPD, 1.63 ± 0.07 versus 1.00; DWF4, 2.76 ± 0.08 versus 1.00) or with a treatment of 100 nM of BL for 2.5 hours (CPD, 0.83 ± 0.05 versus 0.47 ± 0.03; DWF, 0.80 ± 0.06 versus 0.27 ± 0.01). In contrast, the expression of Saur-AC1 was significantly lower in the BKI1-YFP line than wild type in both conditions (no BL, 0.24 ± 0.03 versus 1.00; plus BL, 1.65 ± 0.11 versus 4.68 ± 0.31). Together with the hypocotyl growth results, these results support the interpretation that BKI1 is a negative regulator of brassinosteroid signaling.

To investigate where in the pathway BKI1 acts, we examined the subcellular localization of BKI1-YFP fusion protein in plants. BKI1-YFP was localized to both the plasma membrane and the cytosol in root-tip cells (Fig. 3A). When the BKI1-YFP line was grown on medium containing 1 μM BRZ220, which reduces the levels of endogenous brassinosteroids (22), the fluorescent signal on the cell surface was enhanced (Fig. 3B). In contrast, when grown on medium containing 1 μM BL, the plasma membrane association of BKI1-YFP was reduced, as indicated by the diminished discrete fluorescent signal on the cell surface (Fig. 3C), suggesting that brassinosteroids might alter the subcellular localization of BKI1-YFP. After BL treatment, we observed that the plasma membrane association of BKI1-YFP was almost completely gone after 5 min and undetectable 10 min later (Fig. 3, G to I, and movie S2); in contrast, the BKI1-YFP signal was unchanged in the control (Fig. 3, D to F, and movie S1).

Fig. 3.

Brassinosteroid treatment triggers the dissociation of BKI1-YFP from the plasma membrane in a BRI1-dependent manner. (A) BKI1-YFP is localized to the plasma membrane and cytosol on Murashige and Skoog (MS) medium. (B) BRZ220 enhances the association of BKI1-YFP to the plasma membrane. Seedlings were grown on MS medium containing 1 μM BRZ220. (C) BL enhances BKI1-YFP dissociation from the plasma membrane. Seedlings were grown on MS medium containing 1 μM BL. (D to F) Dimethylsulfoxide (DMSO) at 0.001% (v/v) does not alter the plasma membrane localization of BKI1-YFP. (G to I) BL-induced dissociation of BKI1-YFP from the plasma membrane. Seedlings were grown on MS medium containing 1 μM of BRZ220 and treated with 1 μM BL. (J to L) BKI1-YFP is associated with the plasma membrane in bri1-116. Seedlings were grown on MS medium containing 1 μM of BRZ220. bri1-116 is a null allele of BRI1 (21). (M to O) BL does not cause dissociation of BKI1-YFP from the plasma membrane in bri1-116. Seedlings were grown on MS medium containing 1 μM BL.(P to R) An active kinase is required for BL-induced dissociation of BKI1 from the plasma membrane. bri1-104 is a loss-of-function mutation in the kinase domain of BRI1 (21). Seedlings were grown on MS medium containing 1μM BRZ220. (D), (G), (J), (M), and (P) were untreated; (E), (H), (K), (N), and (Q) were treated with BL for 5 min; and (F), (I), (L), (O), and (R) were treated with BL for 10 min.

To assess whether the plasma membrane association of BKI1-YFP was dependent on BRI1, we introduced the BKI1-YFP construct into a bri1-116 background, a null allele of bri1 that generates a stop codon within the BL binding domain (21). In bri1-116 homozygous seedlings, we observed that a substantial amount of BKI1-YFP was localized to the cell surface. Unlike wild-type plants, BL treatment failed to induce the dissociation of BRI1 (Fig. 3, J to O), indicating that BRI1 is not required for the plasma membrane association of BKI1-YFP but is crucial for the release of BKI1 from the plasma membrane. Likewise, in bri1-104, a kinase-inactive allele of bri1 (21) that accumulates normal levels of protein, no BL-induced dissociation of BKI1-YFP from the plasma membrane was observed, suggesting that the kinase activity of BRI1 is required for the dissociation of BKI1-YFP from the plasma membrane (Fig. 3, P to R). These results imply that BL-induced phosphorylation of BKI1 by activated BRI1 is essential for regulating the subcellular localization and association of BKI1 with BRI1. In corroboration, we found that the yeast two-hybrid interaction of BKI1 with a kinase-inactive BRI1 was stronger than with the kinase-active BRI1, although the kinase-active version could still interact with BKI1 (fig. S2). In addition, BKI1 is a substrate of BRI1 kinase in vitro with a Michaelis constant (Km) of about 1.55 μM (fig. S3A) and is a phosphoprotein in vivo as indicated by a shift following λ-phosphatase treatment (fig. S3B).

To confirm whether the plasma membrane association of BKI1 is required for it to inhibit BRI1 signaling, we made transgenic plants harboring a BKI1-YFP fusion protein that was tethered to the plasma membrane by adding an N-terminal myristoylation site (myriBKI1-YFP). As shown in Fig. 4A, the myriBKI1-YFP was constitutively associated with the plasma membrane even after BL treatment. As expected for a regulator whose inhibitory function is predicted to require an interaction with BRI1 at the plasma membrane, the myriBKI1-YFP line had an enhanced dwarf phenotype when compared with plants expressing BKI1-YFP to a similar level (Fig. 4B).

Fig. 4.

A model for BKI1 in BRI1 activation. (A) A myristoylated BKI1-YFP is constitutively associated with the plasma membrane following BL treatment. Seedlings were grown on MS medium containing 1 μM of BRZ220. The time of BL treatment is indicated. Upper panels, BKI1-YFP; lower panels, myriBKI1-YFP. (B) Overexpression of myriBKI1-YFP leads to an enhanced dwarf phenotype. Immunoblots with antibody to green fluorescent protein (GFP) show levels of myriBKI1-YFP and BKI1-YFP. Bottom band, loading control. (C) BKI1 inhibits the interaction between the kinase domains of BRI1 and BAK1 in vitro. The total 35S-Met–labeled BRI1-KD-HIS coprecipitated by GST-BAK1-KD was defined as “1.” Five replicates were conducted. Error bars indicate standard error. 1, GST-BAK1-KD + 35S-Met-labeled BRI1-KD-HIS; 2, GST-BAK1-KD + 35S-Met-labeled BRI1-KD-HIS + 10 μM MBP; 3, GST-BAK1-KD + 35S-Met-labeled BRI1-KD-HIS + 10 μM MBP-BKI1. The bottom panel shows a representative gel by autoradiography, indicating the pull-down 35S-Met labeled BRI1-KD-HIS. (D) A model to illustrate the role of BKI1 in BRI1 signaling. Without BL, BRI1 kinase is kept in a basal state by both its own carboxyl terminal domain and by an interaction with BKI1. Brassinosteroid binding to the extracellular domain of BRI1 induces a conformational change of the kinase domain, leading to the phosphorylation of the C-terminal domain of BRI1 and BKI1, the dissociation of BKI1 from the plasma membrane, and the release of autoinhibition of BRI1. These events lead to the full activation of BRI1 and its association with BAK1 or other substrates. Plasma membrane–dissociated BKI1 may also regulate other unknown components in the brassinosteroid signaling cascade. BSU1 is a nuclear serine/threonine phosphatase involved in the dephosphorylation of BES1 (27).

Our results suggest that BKI1 may function early in the brassinosteroid signaling pathway, either as an early signaling component or as a regulator that inactivates or desensitizes the receptor. We looked at the interaction of BKI1 with BIN2, the earliest known signaling component downstream of the brassinosteroid receptor complex, but were unable to detect any interaction using genetic assays in yeast (Fig. 1D). Because BAK1 interactions with BRI1 are more stable after application of brassinosteroids (11), we speculated that BKI1 may inhibit the association of BRI1 with positive regulators, e.g., BAK1. To test this prediction, we conducted GST pull-down experiments in vitro and observed that the addition of a BKI1-maltose binding protein fusion (MBP-BKI1) significantly inhibited (by ∼62%, P = 0.000013) the interaction between the kinase domains of BRI1 and BAK1, whereas the addition of MBP alone did not significantly affect their interaction (P = 0.219) (Fig. 4C).

The simplest interpretation of our observations is that plasma membrane–associated BKI1 interacts directly with BRI1 and represses its signaling (Fig. 4D). This model predicts that, in the absence of steroid, BKI1 is localized to the plasma membrane, where it binds to the intracellular domain of a BRI1 homodimer, thus keeping BRI1 from associating with BAK1. Brassinosteroid binding to the extracellular domain of BRI1 induces receptor phosphorylation and activation, as well as its dissociation from BKI1, leaving open binding sites for BAK1 and perhaps other BRI1 substrates. BKI1 binding to BRI1 maintains a low basal activity of BRI1. This low activity allows expression of brassinosteroid biosynthetic genes, which are repressed by BRI1 signaling (16, 23) (Fig. 2J). BKI1's role in BRI1 signaling thus appears to be distinct from that of TTL, a negative regulator of the brassinosteroid signal transduction pathway, which associates with a kinase-active rather than an inactive BRI1 kinase (18).

In metazoans, adaptor proteins associate with receptor tyrosine kinases and recruit signaling components to activated receptors (2426). BKI1 does not appear to be a typical adaptor, because it must dissociate from the plasma membrane for BRI1 to signal. Rather, BKI1's association with BRI1's kinase domain prevents it from becoming fully activated and may ensure the specificity of BRI1 signaling.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

Table S1

Movies S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article